https://www.orchest.io/ logo
#tech-support
Title
# tech-support
c

Châu Phước Nguyễn

11/14/2022, 5:10 PM
I just started Orchest on kubectl and got this flashing screen for Jupyter Lab, do you guys have any idea why? Thanks
y

Yannick

11/14/2022, 5:18 PM
@Rick Lamers / @Rasmus Do you have any idea? Could it be something due to the iframe?
r

Rick Lamers

11/14/2022, 5:59 PM
This looks like the <iframe> is pointing to the JupyterLab service that is not running and hence the <iframe> gets redirected to the Orchest application root (projects page). I see you're running this on your own k8s cluster so it's worth it seeing what the
jupyter-server
container says when you start the pipeline session (that happens when you open the pipeline and after that session starts the JupyterLab tab starts working). Likely that container is failing to start in your k8s environment. (Attached image: when that checkbox is turned on the session starts and the
jupyter-server
container is started)
c

Châu Phước Nguyễn

11/14/2022, 6:37 PM
I can start the pipeline session
Another thing I notice that an Environment-Shell service is failing, not sure if that is the cause
From f12
r

Rick Lamers

11/14/2022, 6:44 PM
Could you show the network tab to see which URL the <iframe> is trying to load? My guess would be that it might be pointing to an unreachable URL. Likely because you're running on a custom port (:1606)
c

Châu Phước Nguyễn

11/14/2022, 6:47 PM
image.png,image.png
I can just get data from that endpoint directly by the browser or postman. So maybe a CORS thing?
Otherwise maybe frontend issue. Have you tried the latest version from your end?
r

Rick Lamers

11/15/2022, 9:26 AM
to see which URL the <iframe> is trying to load
This is the key, the main application supports running on a different port but I think the <iframe> loaded URL is ignoring the custom port.
Can you show the network tab right after you try to load JupyterLab?
c

Châu Phước Nguyễn

11/15/2022, 9:39 AM
Yes, it's the screenshot above
r

Rick Lamers

11/15/2022, 9:42 AM
Could you set the following filters?
And capture an attempted load of JupyterLab, clicking on the JupyterLab tab (all the way from reloading the page with Cmd+R)